mlsanityData to 14 Oct 2026

Raising the Bar of AI-generated Image Detection with CLIP

arXiv 2312.00195 ↗First posted 30 Nov 2023cs.CVcited by 164

On the map: Multimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP

Abstract (arXiv)

The aim of this work is to explore the potential of pre-trained vision-language models (VLMs) for universal detection of AI-generated images. We develop a lightweight detection strategy based on CLIP features and study its performance in a wide variety of challenging scenarios. We find that, contrary to previous beliefs, it is neither necessary nor convenient to use a large domain-specific dataset for training. On the contrary, by using only a handful of example images from a single generative model, a CLIP-based detector exhibits surprising generalization ability and high robustness across different architectures, including recent commercial tools such as Dalle-3, Midjourney v5, and Firefly. We match the state-of-the-art (SoTA) on in-distribution data and significantly improve upon it in terms of generalization to out-of-distribution data (+6% AUC) and robustness to impaired/laundered data (+13%). Our project is available at https://grip-unina.github.io/ClipBased-SyntheticImageDetection/

Key facts. Cited by 164 papers in the map (data through 14 Oct 2026). Placed in Prompt learning and adaptation of CLIP, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 63 of its 87 references that are in the map. First posted on arXiv 30 Nov 2023.

Builds on

63 of 87 references.

Closest ancestors · 3

Same direction, newest first

  1. 2022Tip-Adapter: Training-free Adaption of CLIP for Few-shot Classification19 Jul
  2. 2021A Simple Baseline for Open-Vocabulary Semantic Segmentation with Pre-trained Vision-language Model29 Dec
  3. 2021Learning to Prompt for Vision-Language Models2 Sep

Other references · 60

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  3. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  4. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  5. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  6. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,255
  7. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,585
  8. 2022Scalable Diffusion Models with Transformerscited by 6,054
  9. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 6,013
  10. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,356
  11. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,322
  12. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,152
38 more
  1. 2021GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Modelscited by 3,614
  2. 2019Analyzing and Improving the Image Quality of StyleGANcited by 3,588
  3. 2020Taming Transformers for High-Resolution Image Synthesiscited by 3,253
  4. 2017Progressive Growing of GANs for Improved Quality, Stability, and Variationcited by 2,988
  5. 2018Large Scale GAN Training for High Fidelity Natural Image Synthesiscited by 2,626
  6. 2021LAION-400M: Open Dataset of CLIP-Filtered 400 Million Image-Text Pairscited by 1,555
  7. 2015LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loopcited by 1,355
  8. 2021Efficient Geometry-aware 3D Generative Adversarial Networkscited by 1,205
  9. 2021Alias-Free Generative Adversarial Networkscited by 1,184
  10. 2022eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoiserscited by 931
  11. 2019CNN-generated images are surprisingly easy to spot... for nowcited by 847
  12. 2023DataComp: In search of the next generation of multimodal datasetscited by 644
  13. 2020Leveraging Frequency Analysis for Deep Fake Image Recognitioncited by 504
  14. 2023Scaling up GANs for Text-to-Image Synthesiscited by 492
  15. 2023Towards Universal Fake Image Detectors that Generalize Across Generative Modelscited by 490
  16. 2023DIRE for Diffusion-Generated Image Detectioncited by 390
  17. 2019Detecting and Simulating Artifacts in GAN Fake Imagescited by 293
  18. 2022On the detection of synthetic images generated by diffusion modelscited by 283
  19. 2020Watch your Up-Convolution: CNN Based Generative Deep Neural Networks are Failing to Reproduce Spectral Distributionscited by 280
  20. 2020What makes fake images detectable? Understanding properties that generalizecited by 273
  21. 2022Delving into Out-of-Distribution Detection with Vision-Language Representationscited by 216
  22. 2023StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesiscited by 212
  23. 2022Diffusion-GAN: Training GANs with Diffusioncited by 205
  24. 2018Do GANs leave artificial fingerprints?cited by 190
  25. 2022ZegCLIP: Towards Adapting CLIP for Zero-shot Semantic Segmentationcited by 178
  26. 2022DE-FAKE: Detection and Attribution of Fake Images Generated by Text-to-Image Generation Modelscited by 168
  27. 2021Zero-Shot Out-of-Distribution Detection Based on the Pre-trained Model CLIPcited by 152
  28. 2019Fourier Spectrum Discrepancies in Deep Network Generated Imagescited by 139
  29. 2021Are GAN generated images easy to detect? A critical analysis of the state-of-the-artcited by 139
  30. 2023Intriguing properties of synthetic images: from generative adversarial networks to diffusion modelscited by 123
  31. 2022Diffusion Probabilistic Model Made Slimcited by 121
  32. 2018ForensicTransfer: Weakly-supervised Domain Adaptation for Forgery Detectioncited by 104
  33. 2023Art and the science of generative AI: A deeper divecited by 95
  34. 2023Identifying and Mitigating the Security Risks of Generative AIcited by 91
  35. 2022Fusing Global and Local Features for Generalized AI-Synthesized Image Detectioncited by 88
  36. 2023GALIP: Generative Adversarial CLIPs for Text-to-Image Synthesiscited by 85
  37. 2021A Closer Look at Fourier Spectrum Discrepancies for CNN-generated Images Detectioncited by 68
  38. 2019Incremental learning for the detection and classification of GAN-generated imagescited by 58

50 most cited of 60 All 60 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 43; 2025: 59; 2026: 62.

164 papers.

Sort:

By area:

  1. 2024Detecting Multimedia Generated by Large AI Models: A Surveycited by 76
  2. 2024A Bias-Free Training Paradigm for More General AI-generated Image Detectioncited by 67
  3. 2025Dual Data Alignment Makes AI-Generated Image Detector Easier Generalizablecited by 60
  4. 2024Any-Resolution AI-Generated Image Detection by Spectral Learningcited by 57
  5. 2024Zero-Shot Detection of AI-Generated Imagescited by 50
  6. 2024Community Forensics: Using Thousands of Generators to Train Fake Image Detectorscited by 50
  7. 2024CLIPping the Deception: Adapting Vision-Language Models for Universal Deepfake Detectioncited by 47
  8. 2024D3: Scaling Up Deepfake Detection by Learning from Discrepancycited by 45
  9. 2024On Learning Multi-Modal Forgery Representation for Diffusion Generated Video Detectioncited by 40
  10. 2025Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detectorcited by 40
  11. 2024FakeBench: Probing Explainable Fake Image Detection via Large Multimodal Modelscited by 36
  12. 2024Aligned Datasets Improve Detection of Latent Diffusion-Generated Imagescited by 36
38 more
  1. 2025AIGI-Holmes: Towards Explainable and Generalizable AI-Generated Image Detection via Multimodal Large Language Modelscited by 33
  2. 2024Standing on the Shoulders of Giants: Reprogramming Visual-Language Model for General Deepfake Detectioncited by 30
  3. 2024AI-Face: A Million-Scale Demographically Annotated AI-Generated Face Dataset and Fairness Benchmarkcited by 28
  4. 2025Bridging the Gap Between Ideal and Real-world Evaluation: Benchmarking AI-Generated Image Detection in Challenging Scenarioscited by 24
  5. 2024Circumventing shortcuts in audio-visual deepfake detection datasets with unsupervised learningcited by 23
  6. 2024FSFM: A Generalizable Face Security Foundation Model via Self-Supervised Facial Representation Learningcited by 23
  7. 2025Seeing What Matters: Generalizable AI-generated Video Detection with Forensic-Oriented Augmentationcited by 23
  8. 2024Organic or Diffused: Can We Distinguish Human Art from AI-generated Images?cited by 21
  9. 2024Bi-LORA: A Vision-Language Approach for Synthetic Image Detectioncited by 20
  10. 2024Robust CLIP-Based Detector for Exposing Diffusion Model-Generated Imagescited by 20
  11. 2025Toward Generalizable Forgery Detection and Reasoningcited by 20
  12. 2024DeCLIP: Decoding CLIP representations for deepfake localizationcited by 17
  13. 2024Are CLIP features all you need for Universal Synthetic Image Origin Attribution?cited by 14
  14. 2025Characterizing Photorealism and Artifacts in Diffusion Model-Generated Imagescited by 14
  15. 2026DNA: Uncovering Universal Latent Forgery Knowledgecited by 14
  16. 2024Synthetic Image Verification in the Era of Generative AI: What Works and What Isn't There Yetcited by 13
  17. 2024Vulnerabilities in AI-generated Image Detection: The Challenge of Adversarial Attackscited by 12
  18. 2025Methods and Trends in Detecting AI-Generated Images: A Comprehensive Reviewcited by 12
  19. 2024Exploring the Adversarial Robustness of CLIP for AI-generated Image Detectioncited by 11
  20. 2025Stay-Positive: A Case for Ignoring Real Image Features in Fake Image Detectioncited by 11
  21. 2025Forensic Self-Descriptions Are All You Need for Zero-Shot Detection, Open-Set Source Attribution, and Clustering of AI-generated Imagescited by 11
  22. 2025FakeScope: Large Multimodal Expert Model for Transparent AI-Generated Image Forensicscited by 11
  23. 2025Manifold Induced Biases for Zero-shot and Few-shot Detection of Generated Imagescited by 10
  24. 2025Navigating the Challenges of AI-Generated Image Detection in the Wild: What Truly Matters?cited by 10
  25. 2025OpenFake: An Open Dataset and Platform Toward Real-World Deepfake Detectioncited by 10
  26. 2025Training-free Detection of AI-generated images via Cropping Robustnesscited by 10
  27. 2024TextureCrop: Enhancing Synthetic Image Detection through Texture-based Croppingcited by 9
  28. 2025WILD: a new in-the-Wild Image Linkage Dataset for synthetic image attributioncited by 9
  29. 2024Training-Free Deepfake Voice Recognition by Leveraging Large-Scale Pre-Trained Modelscited by 8
  30. 2024Evolution of Detection Performance throughout the Online Lifespan of Synthetic Imagescited by 8
  31. 2024Wavelet-Driven Generalizable Framework for Deepfake Face Forgery Detectioncited by 8
  32. 2025TrueFake: A Real World Case Dataset of Last Generation Fake Images also Shared on Social Networkscited by 8
  33. 2025Rethinking Cross-Generator Image Forgery Detection through DINOv3cited by 8
  34. 2026Training-free Detection of Generated Videos via Spatial-Temporal Likelihoodscited by 8
  35. 2026Venus-DeFakerOne: Unified Fake Image Detection & Localizationcited by 8
  36. 2024Detecting Origin Attribution for Text-to-Image Diffusion Modelscited by 7
  37. 2024When Synthetic Traces Hide Real Content: Analysis of Stable Diffusion Image Launderingcited by 7
  38. 2024Adversarial Robustness of AI-Generated Image Detectors in the Real Worldcited by 7

50 most cited of 164 All 164 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025NS-Net: Decoupling CLIP Semantic Information through NULL-Space for Generalizable AI-Generated Image Detection
  2. 2023Alpha-CLIP: A CLIP Model Focusing on Wherever You Want
  3. 2024Mixture of Low-rank Experts for Transferable AI-Generated Image Detection
  4. 2023CLIP meets Model Zoo Experts: Pseudo-Supervision for Visual Enhancement
  5. 2025An Empirical Analysis of VLM-based OOD Detection: Mechanisms, Advantages, and Sensitivity
  6. 2024Toward a Holistic Evaluation of Robustness in CLIP Models
  7. 2023Improving CLIP Robustness with Knowledge Distillation and Self-Training
  8. 2025FA: Forced Prompt Learning of Vision-Language Models for Out-of-Distribution Detection